Radio Galaxy Zoo Talk

Artefacts and coordinates

  • sisifolibre by sisifolibre

    All the objects in FIRST that haven't counterpart in NVSS can be cosidered artefacts for RGZ?. Assuming yes, ARG00022ol is an artefact, and I have used the list of subjects (how do I find ARG fields) to search the continuation of the stripes that I see in FIRST. For this reason, I think that ARG00022oo and ARG00022op are artefacts too.

    I have made a search using hastag #artefact and I have found that ARG00022ps can be part of the same artefact, and for extension ARG00022pr.

    Can the staff or sciencist remove subjects for the classification?. If the answer is yes, I think that you would save many clicks if you remove artifacts.

    Posted

  • JeanTate by JeanTate in response to sisifolibre's comment.

    All the objects in FIRST that haven't counterpart in NVSS can be considered artefacts for RGZ?

    In a word, no. I'll see if I can find some ARG examples, later, but a quick look at what the two surveys are hints at why there can be real FIRST sources which are not NVSS ones ...

    From this NVSS website (my bold):

    The principal data products of the NVSS were [...] plus (2) a catalog of almost 2 million discrete sources stronger than a flux density S of about 2.5 mJy. The images all have 45 arcsecond FWHM angular resolution and nearly uniform sensitivity.

    And from this FIRST website:

    A final atlas of maps is produced by coadding the twelve images adjacent to each pointing center. These maps have 1.8" pixels, a typical rms of 0.15 mJy, and a resolution of 5". The noise in the coadded maps varies by only 15% from the best to the worst places in the maps, except in the vicinity of bright sources (> 100 mJy) where sidelobes can lead to an increased noise level. At the 1 mJy source detection threshold, there are ~90 sources per square degree, ~35% of which have resolved structure on scales from 2-30". A source catalog including peak and integrated flux densities and sizes derived from fitting a two-dimensional Gaussian to each source is generated from the coadded images. The astrometric reference frame of the maps is accurate to 0.05", and individual sources have 90% confidence error circles of radius < 0.5" at the 3 mJy level and 1" at the survey threshold.

    FIRST has both a better resolution (5") and lower source detection threshold (1 mJy) than NVSS (45" and 2.5 mJy); on the other hand, it seems that NVSS has more uniform sensitivity. One place to start checking on whether a FIRST source is an artifact is the FIRST catalog; in particular, check out the P(S) value:

    P(S) indicates the probability that the source is spurious (most commonly because it is a sidelobe of a nearby bright source.)

    As I think 42jkb said - somewhere - RGZ did NOT^ use P(S) in its selection of FIRST sources to include in ARG fields, so even if a FIRST source has a P(S) value of, say, 0.9, we will get the field centered on it to classify.

    However, one possible reason for why there would be a real FIRST source but not an NVSS one does not apply: NVSS has a greater areal (on the sky) coverage than FIRST does, and it includes all FIRST fields; so there should be no part of the sky with a FIRST source but which is not covered by NVSS.

    Hope this helps! 😃

    ^ETA: I mis-spoke; RGZ does/did not consider P(S) when selecting objects for ARG fields (sorry)

    Posted

  • JeanTate by JeanTate in response to JeanTate's comment.

    I'll see if I can find some ARG examples, later

    Hmm ... not as easy as I had expected ...

    Anyway, the host of the FIRST source in ARG0002359 - which is overedge - is SDSS J151353.96+241232.1 aka 2MASX J15135400+2412317 (thank you Dolorous Edd). The doublelobe nature of this source is obvious in NVSS:

    enter image description here

    However, the NE lobe is missing in FIRST ... however, the core is obvious in FIRST, but missing in NVSS:

    enter image description here

    Of course, this may not be the kind of thing you were referring to, in the OP ...

    Posted

  • sisifolibre by sisifolibre

    Jean Thank you very much for your detailed explanation, you're a great teacher!. 😃

    You answer one of the questions posed in my original post. About the other question... It is possible that I am sometimes a little ingenuous asking and making suggestions 😛 . Now, I guess that the staff of RGZ already have considered if worth remove subjects that correspond to artefacts and others like those hidden by stars and similar.

    Posted

  • sisifolibre by sisifolibre

    FIRST has both a better resolution (5") and lower source detection threshold (1 mJy) than NVSS (45" and 2.5 mJy).

    Then we can say that objects in FIRST with brightest pixel is higher than 2.5 mJy and don't have counterpart in NVSS, would be artefacts?. ARG00022ol indicates 3.42 mJy, is bigger than 45'' and it doesn't appear in NVSS.

    I want to clarify that when I say #artefact I refer that perhaps the signal does not come from a radio galaxy, not necessarily that is some kind of error in the detection of the signal. I don't know if there are other objects that can appear in these FIRST images...

    I can't find P(S) meaning and where to check out his value 😦. Excuse my, I'm learning very slowly, step by step, reading articles about my doubts 😄

    Posted

  • JeanTate by JeanTate in response to sisifolibre's comment.

    Then we can say that objects in FIRST with brightest pixel is higher than 2.5 mJy and don't have counterpart in NVSS, would be artefacts?

    Likely, yes; certainly? no. Sensitivity is also important, and while it's pretty uniform in NVSS it's not 100% so (there may be parts of the sky where the NVSS sensitivity is low, so sources of ~2.5 mJy would not be detected).

    I want to clarify that when I say #artefact I refer that perhaps the signal does not come from a radio galaxy, not necessarily that is some kind of error in the detection of the signal. I don't know if there are other objects that can appear in these FIRST images...

    Radio astronomy is not like optical astronomy, at least the interferometry part of it. Some features appear somewhat similar, for example in optical images you often get four 'compass points' with bright stars (technical term is 'diffraction spikes'), and in radio astronomy the hexagon-pattern of artifacts around bright (point-like) radio sources. However, the artifacts which are 'side lobes' in radio astronomy may appear quite some distance (on the sky) away from the source which produces them; in this sense they're perhaps a bit more like 'ghosts' in optical astronomy (how they come to be is, of course, very different).

    I can't find P(S) meaning and where to check out his value

    I don't know what radio astronomers do, but I have downloaded the FIRST catalog, and regularly use it. If you'd like, I can look up some FIRST sources you're interested in, and tell you what the catalog entries for those sources are. Also, I think there is at least one online service where you can look up individual entries in an astronomical catalog; shall I suggest one/some for you?

    Posted

  • sisifolibre by sisifolibre in response to JeanTate's comment.

    No problem, finally I have found it. I was searching in the "image cutouts", but now I see that the P(s) (and his definition) appear in the "catalog search". Thank you.

    Posted

  • KWillett by KWillett scientist, admin, translator

    Just saying that this is a useful discussion - we've used input from users before in retiring subjects early (for example, the faint Hubble blobs #fhb) in the GZ CANDELS images. If we do come up with an accurate method for identifying subjects that should be retired, it'll be seriously considered!

    Posted

  • sisifolibre by sisifolibre

    We can try it. I am very inexpert but I dare to continue the discussion with hope that it helps.
    First, What subjects could be retired? IMHO:

    1: subjects whose central object are artefacts. We can define it using the criteria described in this discussion (latter we can define it better ), and by agreeing to use the hashtag #artefact or #artifact.

    2: subjects clarely unclassifiable. If the classification for the main project in RGZ can be only by FIRST, then may be that some subjects whose hosts are "hidden" by a bigger IR source like #stars or near galaxys, can be retired to the classification.

    Posted

  • JeanTate by JeanTate in response to sisifolibre's comment.

    Cool! 😃 Despite what you say, sisifolibre, I think that you are, by now, very far from being inexpert!

    I agree with this, as a criterion for selecting fields for consideration as ones to be retired:

    subjects whose central object are artefacts. We can define it using the criteria described in this discussion (latter we can define it better ), and by agreeing to use the hashtag #artefact or #artifact.

    However, I think each such field so tagged would need to be looked at, to see if the tag refers to (all) the radio source(s), or just one (some) of them ... for example, there are fields full of artifacts, but which also contain 'real' radio sources (e.g. those with a bright, central radio source, which causes many/most/all of the artifacts in that field).

    Posted

  • sisifolibre by sisifolibre

    Ok, then if scientists agree, #stars and similar can't be removed, and we consider to remove only #arti(e)fact(s).

    I agree with that each field tagged must be reviewed, and this tags can't be a criterion, they are only a first selection.

    I'm not sure but AFAIK each radio object have a ARG subject, then fields with lots of artefacts (like caused by a QSO) have lots of ARG fields (although I have not found many of these). If we find one of these subjects with lots of artefacts and we suspect that one of them are a real source, we can search the ARG subjects that have the real source in the center using the list of subjects, and remove the others. Other way can be search in other catalogs the same field.

    The criterion on simpler objects can begin with: objects in FIRST with brightest pixel is higher than 2.5 mJy, don't have counterpart in NVSS, and don't have an, IR or optical, evident host.

    If @KWiillett think that worth, we can begin making a new thread where see and discuss objects that are tagged like #arte(i)fact(s), searching in the ARG list the related subjects, and look some of the already tagged. Or can be too much work?

    Posted

  • KWillett by KWillett scientist, admin, translator

    Gah. Hate it when my browser crashes and I have to rewrite the post.

    So - I think this is a great discussion, @sisifolibre and @JeanTate. I think the most useful thing for this stage of the project would be to identify patterns in artifacts that people have already tagged in RGZ. We've built up a good-sized sample at this point, and that would let us do some analysis on what the commonalities are between them. If we can identify parameters that all artifacts have in common (and yet aren't seen in the real radio galaxy images), we could then check the remaining ~50,000 images in FIRST and ATLAS to remove potential artifacts and thus have all of our volunteers spend their time more effectively, classifying real images.

    As a rough process, we might try:

    • assembling a collection of tagged artifacts
    • looking at (at least some of them) by eye
    • discuss which parameters we could automatically measure
    • pick values of some parameters (eg, size, total flux density in the image, shape, ...) that are common to all artifacts
    • see how many other, unseen images in RGZ have the same values and check by eye if they also appear to be artifacts
    • remove the new artifacts from RGZ

    If you're interested in helping, that'd be tremendous. For the first step - you can use the built-in Search function in Talk to find objects or collections that are tagged as artifacts. However, I know the Search function can be fincky sometimes, and the paginated results may not be the most useful. Alternatively, here's a list of the IDs for every object tagged as #artifact(s) or #artefact(s) on RGZ so far. You can then find objects manually using the URL (for example, http://radiotalk.galaxyzoo.org/#/subjects/ARG0002poh).

    https://www.dropbox.com/s/ikiwfmsrlz5qzfx/artifacts_tagged.csv?dl=0

    I'd suggest we could start by looking at the tagged images w/artifacts and making a list of the reasons why they're being tagged that way (eg, what users are seeing in the radio/IR). If we come up with some good ideas, I can help supply more data and we'll explore whether this is something we can reliably detect.

    Posted

  • sisifolibre by sisifolibre

    Of course I'm interested, but I just hope that my contributions are helpful in this. Please, If I say or make something wrong or anyone have a better idea, please tell me.

    I think we can use the Search function, seems that it work well if you refresh the page before to make a search, and if you repeat a search with the same word it show the same order of subjects.

    I'm going to begin some collections with the subjects tagged as #arte(i)fact(s).I mean, some collections with what I think that are different kinds of artefacts and one collection with all of them (to bring them all and in the darkness bind them 😛). Please, tell me if any of these collections are useless or if any is useful!

    Total collection: All the subgects that have been tagged about possible radio artefacts:
    http://radiotalk.galaxyzoo.org/#/collections/CRGS00017a

    Collection about objects that seem artefactS caused by strong nearby emission (the radio source is alwais out of the field of the subject):
    http://radiotalk.galaxyzoo.org/#/collections/CRGS00017c
    A clue for this kind of artefacts is that they usually are in a "noise line" diagonal to the strong emision. Sometimes the source of strong emision have the jets aligned with the diagonal "noise lines".

    Collection about subjects that have artefacts and real radio source in his fields:
    http://radiotalk.galaxyzoo.org/#/collections/CRGS00017d
    Some of this subjects have in the center the source that causes the artefacts, but others no and the artefacts are the main objects of the field(may be better make 2 collections of this?). http://radiotalk.galaxyzoo.org/#/subjects/ARG00034d4

    Collection about objects that seems like a brand stape or scratch in the paper: http://radiotalk.galaxyzoo.org/#/collections/CRGS00017e this kind of artefact may be caused by a bright near source, but more far, that is in his horizontal (same Dec.), and have a similar shape in all the cases, its easy to see a repetition and pattern in FIRST.

    Collection about subjects that aren't artefacts because have a NVSS counterpart, a scientist have said it, or some similar evidence: http://radiotalk.galaxyzoo.org/#/collections/CRGS00017b

    Collection about subjects that have bad IR image: http://radiotalk.galaxyzoo.org/#/collections/CRGS000188

    Posted

  • sisifolibre by sisifolibre

    A kind of object that I think can be easy to recognise is this. The subjects with this kind of objects have straight contour lines, and sometimes looks like the mark of a stapler in the paper (its hard to me explain this in english).

    http://radiotalk.galaxyzoo.org/#/subjects/ARG0001zp1:

    enter image description here

    http://radiotalk.galaxyzoo.org/#/subjects/ARG00022ol:

    enter image description here

    http://radiotalk.galaxyzoo.org/#/subjects/ARG00022op:

    enter image description here

    http://radiotalk.galaxyzoo.org/#/subjects/ARG00022ps

    enter image description here

    Posted

  • sisifolibre by sisifolibre

    At first look it seems a restarted with host J085906.20+340400.2, but I think this is another "scratch" example:

    http://radiotalk.galaxyzoo.org/#/subjects/ARG0001hot.

    enter image description here

    I think it is an scratch artefact because:

    • His brightest pixel is 3.74 mJy/beam, over the sensibility of NVSS that don't show a counterpart.
    • Regular pattern W-E
    • Upper an lower radio contours in all lobes are straight and aligned.
    • Line of black dots in the same Dec. (zoom out to 40x40 arcmin/image).

    The P(s) have low values, and all these arguments aren't conclusive but I think that all of then together give a high probability that this subject show an artefact.

    And yes, I must begin to search what more parameters are similar in these artefacts...

    Posted

  • ivywong by ivywong scientist, admin

    Just a note that it can sometimes help to cross-identify the same field with NVSS.

    Posted

  • sisifolibre by sisifolibre

    Yes I know, this is what I mean when I say "NVSS don't show a counterpart". Maybe I'm saying it wrong? Is better to say "the object does not appear in NVSS"? sorry, my english is very limited and I must use automatic translator sometimes...

    I have been comparing in FIRST catalog parameters of artefacts with true faint radio sources, and I can't find any evident difference. In other hand I think that if it were possible it would be done...
    If the parameters in FIRST don't give some clue and the shape can change so much, I have no idea about which parameters we could automatically measure 😦. May be the only way is remove it one by one when they appear.

    Posted

  • ivywong by ivywong scientist, admin in response to sisifolibre's comment.

    Oops sorry sisifolibre. I must have misread. Apologies. No worries, you're doing a good job! Thank you very much.

    Posted

  • sisifolibre by sisifolibre

    Thanks Ivywong. I'm waiting for response about if worth to make the collections of artefacts mentioned here. I'm not sure about if they can be useful

    Posted

  • ivywong by ivywong scientist, admin

    Hi @sisifolibre,

    Apologies for seeing your question earlier. I think that a list of artefacts will be useful for the catalogue team so if you are interested in starting a new artefact thread, please do so. While this artefact thread will not be as much fun as the other ones (such as the hymors or green host for example) in terms of science, this will help the team understand our eventual catalogue better.

    Thank you very much again for your help.

    Posted

  • sisifolibre by sisifolibre

    Well, I was asking about if the diferent collections I did are useful, but no worries.

    I thougt this thread could serve to this topic, but ok, I will start a new artefact thread soon. Collect and classify artefacts is funny for me too.

    Posted

  • ivywong by ivywong scientist, admin

    Oops, I got the wrong end of the stick. But yes, everyone's collections is being inspected by our team in Minnesota and different science members are interested in different aspects of this project so any and all work that you and your fellow SuperZooites have contributed to this project is very much appreciated!

    It is useful for us to know whether a subject contains an artefact or a strange object because very few human eyes have inspected the images that you are currently seeing.

    Thank you again !

    Posted